SPSS Tutorial - 19 | Applied Biostatistics | BIO733_Topic099

Duration: 5:40 634 transcript words EN summary

Video summary

In this module, we explore the utility of quantile-quantile (QQ) plots as a fundamental graphical tool in statistics for assessing whether a dataset plausibly originates from a specific theoretical distribution, such as the normal distribution. This assessment is critical because many statistical analyses rely on the assumption that quantitative variables follow a normal probability distribution. A QQ plot functions as a scatter plot that compares two sets of quantiles: the expected quantiles derived from a theoretical normal distribution and the observed values from the actual dataset. By visualizing these points against each other, analysts can quickly determine if their data adheres to normality or if it deviates significantly. To generate this plot in SPSS, users navigate to the Analyze menu, select Descriptive Statistics, and choose either Explore or specifically the QQ Plot option. Within the dialog box, the variable of interest—in this case, age—is moved into the variables list to test its distributional properties. Users can also estimate distribution parameters directly from the data, which reveals key characteristics like the mean and standard deviation. For the provided dataset, the analysis estimates an average age of 50.98 years, representing the location parameter, and a standard deviation of 16.42 years, indicating the scale or spread where observations typically lie around the center. Interpreting the resulting normal QQ plot involves examining the pattern formed by the plotted dots. If the data perfectly follows a normal distribution, the points will align closely along a straight diagonal line, often referred to as a 45-degree reference line, indicating symmetry and a good fit. However, deviations from this straight line suggest departures from normality. Specifically, while points in the middle of the plot may still follow the trend, outliers at the edges or curves away from the line indicate that the data is not perfectly symmetric or bell-shaped. Such patterns provide clear evidence that the variable does not strictly adhere to a normal distribution, which is essential information for deciding whether parametric statistical tests are appropriate. Beyond standard QQ plots, SPSS also offers detrended normal QQ plots through the same command structure, providing additional ways to visualize distributional nuances. The primary takeaway from this tutorial is that the normal QQ plot serves as a powerful diagnostic instrument to verify the assumption of normality before conducting further statistical analyses. By carefully observing how closely the data points cluster around the central line and identifying any significant outliers or curvature at the tails, researchers can make informed decisions about their data's distributional properties. Ultimately, mastering this tool allows statisticians to validate whether a variable like age in a survey truly represents a normal probability distribution or requires alternative analytical approaches.

Read the full video transcript

In this module, we'll learn about quantile quantile plot, their usability in statistics. We will also learn how to draw quantile quantile plot in SPSS, and we will learn how to interpret them. The QQ plot that look like this kind of plot, or a quantile quantile plot, is a graphical tool that help us to assess if a set of data plausibly come from some theoretical distribution, such as normal or other. Because this is one of the major assumption in many sort of statistical analysis that our quantitative variable should come from the normal probability distribution. Hence, right now, we will learn about normal QQ plot that how we can use the quantile quantile plot to determine if a variable has a normal distribution or not. A QQ plot is a scatter plot that is created by plotting two sets of quantiles against each other. One are expected quantiles from the normal distribution, and the other are observed values from the normal distribution. To to plot this kind of quantile plot, we simply go to analyze, descriptive statistics, explore. One can use it from here. And within the plot command, if you click on normality plot with test, you can simply get the quantile plot. The other way is to go to the analyze descriptive and click on QQ plot. In QQ plot, in the variables tab, you bring the variable that you want to observe that what distribution does it follow? Like here, our interest is to look at the variable age. And we also want to see if in our data the variable age follows the normal probability distribution or not. One can also test the other statistical distributions as well. And one can use any of these options to understand what exact distribution age does follow. But right now, we will only look at if age follows the normal probability distribution or not. Once we brought the variable in, we can put for the distribution parameters, and we can click on that estimate from the data. And there are some other options where we simply click okay. As we hit okay, it gives us output in output viewer. The tool that gives us the information that what's the estimated distribution parameters. As we have discussed previously that the normal probability distribution has two parameters. The first is average, that is the location parameter, and it shows that the average age of the respondents in our data is 50.98 years. And the second is standard deviation, which is a scale parameter. It shows that on average, the observations are 16.42 years away from 50.98, which is the center of the data. So, these values are estimated from our variable age. And the next plot that's given to us is normal QQ plot. In the normal QQ plot, we look at how these dots are being displayed on this scatter. If these dots make a very very good perfect straight line, or they lie right on this line in the middle, which is apparently a 45° line, then we say that our data is a good fit to the normal probability distribution. Or it is symmetric. But this kind of points at the at the edges will show that there are some outlying observations. As well as this indicate that it is not perfectly symmetric. Since we know normal distribution is a pre is a symmetric bell-shaped curve. The presence of these values will give us the hint that no, it is not perfectly bell-shaped. Hence, this is how we interpret the normal quantile-quantile plot, and we use normal quantile-quantile plot to determine that if the distribution of the age is the normal or not. We can also get detrended normal QQ plot as well. Using the same command. Thank you.